Exactly. So the argument comes down to what has great sound vs. what has great playability - and with our current technology, nothing can do both perfectly yet. So you pick what suits you. For me, I don't need to or want to mock up any John Williams or Michael Giacchino; I have no interest in sounding like them. For others that is what they would prefer to sound like, and so they might be more drawn to the sample-modelling stuff.
If you wanted to do John Williams with really great quality sound as well, you'd need to do some serious sampling - it would require an enormous budget, and a large team to prepare the enormous swath of samples, and some serious programming to make it 'work like sample-modelling but sound like real life.' We're getting close now with some libraries beginning to offer various attack-types on notes for sustains, but then they often offer these at a sacrifice of round-robins. This is more because sampling brass is very time-consuming and riddled with challenges - to sample it thoroughly, players need to play an absolute ton of material, and brass players tire quickly. So if you sample 15 pitches of 'sustain' and offer 2x round robins and 3x dynamics, that requires 90 GOOD takes (there may be a few bad.) Now if you offer sustain soft-attack, normale, strong-attack, you're tripling that all. So you cut out the round-robin, maybe even some of the dynamics of certain attacks, to get back down to 90 - which sacrifices some realism and/or flexibility.
Let's not even forget brass players tire quickly. This means they're gonna require breaks, or even for multiple sessions across various days, especially when you get to the really loud stuff.
One day someone might work up the budget and time to do something insane, but for now, Spitfire and Orchestral Tools seem to be on the right path, even if they don't provide everything and the kitchen sink just yet. Play-ability is a whole other ball park and takes a lot of doctoring to get right as well, especially if you want to avoid creating unnatural sounds when blending your recordings.